Machine Learning Human Rights and Wrongs: How the Successes and Failures of Supervised Learning Algorithms Can Inform the Debate About Information Effects
Title: Machine Learning Human Rights and Wrongs: How the Successes and Failures of Supervised Learning Algorithms Can Inform the Debate About Information Effects
There is an ongoing debate about whether human rights standards have changed over the last 30 years. The evidence for or against this shift relies upon indicators created by human coders reading the texts of human rights reports. To help resolve this debate, we suggest translating the question of changing standards into a supervised learning problem. From this perspective, the application of consistent standards over time implies a time-constant mapping from the textual features in reports to the human coded scores. Alternatively, if the meaning of abuses have evolved over time, then the same textual features will be labeled with different numerical scores at distinct times. Of course, while the mapping from natural language to numerical human rights score is a highly complicated function, we show that these two distinct data generation processes imply divergent overall patterns of accuracy when we train a wide variety of algorithms on older versus newer sets of observations to learn how to automatically label texts with scores. Our results are consistent with the expectation that standards of human rights have changed over time. more »« less
PARK, BAEKKWAN; GREENE, KEVIN; COLARESI, MICHAEL(
, American Political Science Review)
null
(Ed.)
This manuscript helps to resolve the ongoing debate concerning the effect of information communication technology on human rights monitoring. We reconceptualize human rights as a taxonomy of nested rights that are judged in textual reports and argue that the increasing density of available information should manifest in deeper taxonomies of human rights. With a new automated system, using supervised learning algorithms, we are able to extract the implicit taxonomies of rights that were judged in texts by the US State Department, Amnesty International, and Human Rights Watch over time. Our analysis provides new, clear evidence of change in the structure of these taxonomies as well as in the attention to specific rights and the sharpness of distinctions between rights. Our findings bridge the natural language processing and human rights communities and allow a deeper understanding of how changes in technology have affected the recording of human rights over time.
Rahman, Safwanur
Hamid(
, IEEE Signal Processing in Medicine and Biology Symposium (SPMB))
Obeid, Iyad; Selesnick, Ivan; Picone, Joseph
(Ed.)
The Temple University Hospital Seizure Detection Corpus (TUSZ) [1] has been in distribution since April 2017. It is a subset of the TUH EEG Corpus (TUEG) [2] and the most frequently requested corpus from our 3,000+ subscribers. It was recently featured as the challenge task in the Neureka 2020 Epilepsy Challenge [3]. A summary of the development of the corpus is shown below in Table 1.
The TUSZ Corpus is a fully annotated corpus, which means every seizure event that occurs within its files has been annotated. The data is selected from TUEG using a screening process that identifies files most likely to contain seizures [1]. Approximately 7% of the TUEG data contains a seizure event, so it is important we triage TUEG for high yield data. One hour of EEG data requires approximately one hour of human labor to complete annotation using the pipeline described below, so it is important from a financial standpoint that we accurately triage data.
A summary of the labels being used to annotate the data is shown in Table 2. Certain standards are put into place to optimize the annotation process while not sacrificing consistency. Due to the nature of EEG recordings, some records start off with a segment of calibration. This portion of the EEG is instantly recognizable and transitions from what resembles lead artifact to a flat line on all the channels. For the sake of seizure annotation, the calibration is ignored, and no time is wasted on it. During the identification of seizure events, a hard “3 second rule” is used to determine whether two events should be combined into a single larger event. This greatly reduces the time that it takes to annotate a file with multiple events occurring in succession. In addition to the required minimum 3 second gap between seizures, part of our standard dictates that no seizure less than 3 seconds be annotated. Although there is no universally accepted definition for how long a seizure must be, we find that it is difficult to discern with confidence between burst suppression or other morphologically similar impressions when the event is only a couple seconds long. This is due to several reasons, the most notable being the lack of evolution which is oftentimes crucial for the determination of a seizure.
After the EEG files have been triaged, a team of annotators at NEDC is provided with the files to begin data annotation. An example of an annotation is shown in Figure 1. A summary of the workflow for our annotation process is shown in Figure 2. Several passes are performed over the data to ensure the annotations are accurate. Each file undergoes three passes to ensure that no seizures were missed or misidentified. The first pass of TUSZ involves identifying which files contain seizures and annotating them using our annotation tool. The time it takes to fully annotate a file can vary drastically depending on the specific characteristics of each file; however, on average a file containing multiple seizures takes 7 minutes to fully annotate. This includes the time that it takes to read the patient report as well as traverse through the entire file.
Once an event has been identified, the start and stop time for the seizure is stored in our annotation tool. This is done on a channel by channel basis resulting in an accurate representation of the seizure spreading across different parts of the brain. Files that do not contain any seizures take approximately 3 minutes to complete. Even though there is no annotation being made, the file is still carefully examined to make sure that nothing was overlooked. In addition to solely scrolling through a file from start to finish, a file is often examined through different lenses. Depending on the situation, low pass filters are used, as well as increasing the amplitude of certain channels. These techniques are never used in isolation and are meant to further increase our confidence that nothing was missed. Once each file in a given set has been looked at once, the annotators start the review process. The reviewer checks a file and comments any changes that they recommend. This takes about 3 minutes per seizure containing file, which is significantly less time than the first pass. After each file has been commented on, the third pass commences. This step takes about 5 minutes per seizure file and requires the reviewer to accept or reject the changes that the second reviewer suggested. Since tangible changes are made to the annotation using the annotation tool, this step takes a bit longer than the previous one. Assuming 18% of the files contain seizures, a set of 1,000 files takes roughly 127 work hours to annotate.
Before an annotator contributes to the data interpretation pipeline, they are trained for several weeks on previous datasets. A new annotator is able to be trained using data that resembles what they would see under normal circumstances. An additional benefit of using released data to train is that it serves as a means of constantly checking our work. If a trainee stumbles across an event that was not previously annotated, it is promptly added, and the data release is updated. It takes about three months to train an annotator to a point where their annotations can be trusted. Even though we carefully screen potential annotators during the hiring process, only about 25% of the annotators we hire survive more than one year doing this work.
To ensure that the annotators are consistent in their annotations, the team conducts an interrater agreement evaluation periodically to ensure that there is a consensus within the team. The annotation standards are discussed in Ochal et al. [4]. An extended discussion of interrater agreement can be found in Shah et al. [5].
The most recent release of TUSZ, v1.5.2, represents our efforts to review the quality of the annotations for two upcoming challenges we hosted: an internal deep learning challenge at IBM [6] and the Neureka 2020 Epilepsy Challenge [3]. One of the biggest changes that was made to the annotations was the imposition of a stricter standard for determining the start and stop time of a seizure. Although evolution is still included in the annotations, the start times were altered to start when the spike-wave pattern becomes distinct as opposed to merely when the signal starts to shift from background. This cuts down on background that was mislabeled as a seizure. For seizure end times, all post ictal slowing that was included was removed.
The recent release of v1.5.2 did not include any additional data files. Two EEG files had been added because, originally, they were corrupted in v1.5.1 but were able to be retrieved and added for the latest release. The progression from v1.5.0 to v1.5.1 and later to v1.5.2, included the re-annotation of all of the EEG files in order to develop a confident dataset regarding seizure identification. Starting with v1.4.0, we have also developed a blind evaluation set that is withheld for use in competitions.
The annotation team is currently working on the next release for TUSZ, v1.6.0, which is expected to occur in August 2020. It will include new data from 2016 to mid-2019. This release will contain 2,296 files from 2016 as well as several thousand files representing the remaining data through mid-2019. In addition to files that were obtained with our standard triaging process, a part of this release consists of EEG files that do not have associated patient reports. Since actual seizure events are in short supply, we are mining a large chunk of data for which we have EEG recordings but no reports. Some of this data contains interesting seizure events collected during long-term EEG sessions or data collected from patients with a history of frequent seizures. It is being mined to increase the number of files in the corpus that have at least one seizure event. We expect v1.6.0 to be released before IEEE SPMB 2020.
The TUAR Corpus is an open-source database that is currently available for use by any registered member of our consortium. To register and receive access, please follow the instructions provided at this web page: https://www.isip.piconepress.com/projects/tuh_eeg/html/downloads.shtml. The data is located here: https://www.isip.piconepress.com/projects/tuh_eeg/downloads/tuh_eeg_artifact/v2.0.0/.
Anson, Ian G; Moskovitz, Cary; Anson, Chris M(
, The journal of writing analytics)
Background: Text recycling (hereafter TR)—the reuse of one’s own textual materials from one document in a new document—is a common but hotly debated and unsettled practice in many academic disciplines, especially in the context of peer-reviewed journal articles. Although several analytic systems have been used to determine replication of text—for example, for purposes of
identifying plagiarism—they do not offer an optimal way to compare documents to determine the nature and extent of TR in order to study and theorize this as a practice in different disciplines. In this article, we first describe TR as a common phenomenon in academic publishing, then explore the challenges associated with trying to study the nature and extent of TR within STEM disciplines. We then describe in detail the complex processes we used to create a system for identifying TR across large corpora of texts, and the sentence-level string-distance lexical methods used to refine and test the system (White & Joy, 2004). The purpose of creating such a system is to identify legitimate cases of TR across large corpora of academic texts in different fields of study, allowing meaningful cross-disciplinary comparisons in future analyses of published work. The findings from such investigations will extend and refine our understanding of discourse practices in academic and scientific settings.
Literature Review: Text-analytic methods have been widely developed and implemented to identify reused textual materials for detecting plagiarism, and there is considerable literature on such methods. (Instead of taking up space detailing this literature, we point readers to several recent reviews: Gupta, 2016; Hiremath & Otari, 2014; and Meuschke & Gipp, 2013). Such methods
include fingerprinting, term occurrence analysis, citation analysis (identifying similarity in references and citations), and stylometry (statistically comparing authors’ writing styles; see Meuschke & Gipp, 2013). Although TR occurs in a wide range of situations, recent debate has focused on recycling from one published research paper to another—particularly in STEM fields (see, for example, Andreescu, 2013; Bouville, 2008; Bretag & Mahmud, 2009; Roig, 2008; Scanlon, 2007). An important step in better understanding the practice is seeing how authors actually recycle material in their published work. Standard methods for detecting plagiarism are not directly suitable for this task, as the objective is not to determine the presence or absence of reuse itself, but to study the types and patterns of reuse, including materials that are syntactically but not substantively distinct—such as “patchwriting” (Howard, 1999). In the present account of our efforts to create a text-analytic system for determining TR, we take a conventional alphabetic approach to text, in part because we did not aim at this stage of our project to analyze non-discursive
text such as images or other media. However, although the project adheres to conventional definitions of text, with a focus on lexical replication, we also subscribe to context-sensitive approaches to text production. The results of applying the system to large corpora of published texts can potentially reveal varieties in the practice of TR as a function of different discourse communities and disciplines. Writers’ decisions within what appear to be canonical genres are contingent, based on adherence to or deviation from existing rules and procedures if and when these actually exist. Our goal is to create a system for analyzing TR in groups of texts produced by the same authors in order to determine the nature and extent of TR, especially across disciplinary areas, without judgment of scholars’ use of the practice.
Ghosh, Indrajeet; Ivler, Matthew; Ramamurthy, Sreenivasan Ramasamy; Roy, Nirmalya(
, 2022 IEEE International Conference on Smart Computing (SMARTCOMP))
Language-guided smart systems can help to design next-generation human-machine interactive applications. The dense text description is one of the research areas where systems learn the semantic knowledge and visual features of each video frame and map them to describe the video's most relevant subjects and events. In this paper, we consider untrimmed sports videos as our case study. Generating dense descriptions in the sports domain to supplement journalistic works without relying on commentators and experts requires more investigation. Motivated by this, we propose an end-to-end automated text-generator, SpecTextor, that learns the semantic features from untrimmed videos of sports games and generates associated descriptive texts. The proposed approach considers the video as a sequence of frames and sequentially generates words. After splitting videos into frames, we use a pre-trained VGG-16 model for feature extraction and encoding the video frames. With these encoded frames, we posit a Long Short-Term Memory (LSTM) based attention-decoder pipeline that leverages soft-attention mechanism to map the semantic features with relevant textual descriptions to generate the explanation of the game. Because developing a comprehensive description of the game warrants training on a set of dense time-stamped captions, we leverage two available public datasets: ActivityNet Captions and Microsoft Video Description. In addition, we utilized two different decoding algorithms: beam search and greedy search and computed two evaluation metrics: BLEU and METEOR scores.
Gouvea, J.(
, Proceedings of the 15th International Conference on Computer-Supported Learning (CSCL))
Education researchers have proposed that qualitative and emerging computational
machine learning (ML) approaches can be productively combined to advance analyses of
student-generated artifacts for evidence of engagement in scientific practices. We applied such a combined approach to written arguments excerpted from university students’ biology laboratory reports. These texts are lengthy and contain multiple different features that could be attended to in analysis. We present two outcomes of this combined analysis that illustrate possible affordances of combined workflows: 1) Comparing ML and human-generated scores allowed us to identify and reanalyze mismatches, increasing our overall confidence in the coding; and 2) ML-identified word clusters allowed us to interpret the overlap in meaning between the original coding scheme and the ML predicted scores, providing insight into which features of students’ writing can be used to differentiate rote from more meaningful engagement in scientific argumentation.
Greene, Kevin T., Park, Baekkwan, and Colaresi, Michael. Machine Learning Human Rights and Wrongs: How the Successes and Failures of Supervised Learning Algorithms Can Inform the Debate About Information Effects. Retrieved from https://par.nsf.gov/biblio/10221832. Political Analysis 27.2 Web. doi:10.1017/pan.2018.11.
Greene, Kevin T., Park, Baekkwan, & Colaresi, Michael. Machine Learning Human Rights and Wrongs: How the Successes and Failures of Supervised Learning Algorithms Can Inform the Debate About Information Effects. Political Analysis, 27 (2). Retrieved from https://par.nsf.gov/biblio/10221832. https://doi.org/10.1017/pan.2018.11
Greene, Kevin T., Park, Baekkwan, and Colaresi, Michael.
"Machine Learning Human Rights and Wrongs: How the Successes and Failures of Supervised Learning Algorithms Can Inform the Debate About Information Effects". Political Analysis 27 (2). Country unknown/Code not available. https://doi.org/10.1017/pan.2018.11.https://par.nsf.gov/biblio/10221832.
@article{osti_10221832,
place = {Country unknown/Code not available},
title = {Machine Learning Human Rights and Wrongs: How the Successes and Failures of Supervised Learning Algorithms Can Inform the Debate About Information Effects},
url = {https://par.nsf.gov/biblio/10221832},
DOI = {10.1017/pan.2018.11},
abstractNote = {There is an ongoing debate about whether human rights standards have changed over the last 30 years. The evidence for or against this shift relies upon indicators created by human coders reading the texts of human rights reports. To help resolve this debate, we suggest translating the question of changing standards into a supervised learning problem. From this perspective, the application of consistent standards over time implies a time-constant mapping from the textual features in reports to the human coded scores. Alternatively, if the meaning of abuses have evolved over time, then the same textual features will be labeled with different numerical scores at distinct times. Of course, while the mapping from natural language to numerical human rights score is a highly complicated function, we show that these two distinct data generation processes imply divergent overall patterns of accuracy when we train a wide variety of algorithms on older versus newer sets of observations to learn how to automatically label texts with scores. Our results are consistent with the expectation that standards of human rights have changed over time.},
journal = {Political Analysis},
volume = {27},
number = {2},
author = {Greene, Kevin T. and Park, Baekkwan and Colaresi, Michael},
editor = {null}
}
Warning: Leaving National Science Foundation Website
You are now leaving the National Science Foundation website to go to a non-government website.
Website:
NSF takes no responsibility for and exercises no control over the views expressed or the accuracy of
the information contained on this site. Also be aware that NSF's privacy policy does not apply to this site.
